Abstract

Info

Title:: Application of LightGBM algorithm in classification of patients with Alzheimer’s disease from structural magnetic resonance images

Author(s):: ZHOU Wen; WANG Yu; LI Changsheng; XIAO Hongbing; XING Suxia; Key Laboratory of Food Safety Big Data Technology, School of Computer and Information Engineering, Beijing Technology and Business University, Beijing 100048, China

Keywords:: ; Keywords: Alzheimer’s disease; LightGBM algorithm; structural magnetic resonance image; brain lesion

Abstract:: The study aims to make better use of computer technology to analyze brain changes in patients with Alzheimer’s disease (AD), and to assist the diagnosis of AD. Herein the structural magnetic resonance images of 116 patients with AD, 116 ones with mild cognitive impairment and 117 normal controls from AD neuroimaging initiative database are pre-processed with spm software and then are investigated by correlation analysis to obtain abnormal brain regions. Subsequently, IBASPM software is used to extract the volume of brain lesion as a feature sample. Finally, LightGBM algorithm is used to classify the feature vectors, and the obtained results are compared with the results of support vector machine and XGBoost algorithms. Experimental results reveal that the accuracy rate of LightGBM algorithm to classify the volume of brain lesion reaches 83%. Among the 3 algorithms discussed in this study, namely LightGBM, support vector machine and XGBoost, LightGBM algorithm has the highest accuracy rate in classification. Therefore, it is effective for the paramedical staffs to perform an early diagnosis of AD using LightGBM algorithm.